Péter Gál

Optimizing coverage in mobile wireless sensor networks

The wireless sensor networks are distributed systems of a special type, mainly due to the limitations of the parts. In most of the realizations the nodes are placed in a grid planned in advance, but in many cases these units are scattered more or less randomly through an area. By giving some nodes the ability to move — with a very simple robot, for example — these systems can autonomously order themselves, by aligning the units by an optimal grid. During the development, the method of potential fields showed the most promising results. In an optimal solution, this results in hexagon grid, with clear radio channels between every node and their six neighbors. The algorithm requires a more-or-less precise localization algorithm. A simulator is also developed to emulate the operation of multiple nodes. This program is based on the built-in simulator library of the TinyOS, and uses its model to create an event-driven simulation and to emulate the behavior of the radio waves. It got extended with the ability to model the position and motion of the nodes. In its current state, the program provides several hooks for extension. Providing artificial errors for the applications are also possible, as the original simulator does so.